In the speech - language pathology research and clinical work, narration is one of the measures of pragmatic abilities. The starting point of this thesis is the analysis on the microstructural level, which was used to determine what the types of words used by children for character reference in narration say about their pragmatic abilities. For this purpose, the words that children use to introduce characters to the narrative and establish anaphoric reference were observed in comparison with the words used by adults. The research involved three groups of subjects: preschool-age children (5;04-6;05 years old), early school-age children (6;06-7;06 years old) and adult speakers as the control group. The storytelling method selected for the purposes of the research is based on the pictorial material from the Croatian version of MAIN: Multilingual Assessment Instrument for Narratives. What was observed is the way in which subjects refer to the five characters in the story Goats. In order to compare the three groups of subjects, a nonparametric statistical analysis was conducted. Statistically significant differences between the two groups of children and the group of adults were found in all of the tested variables: the introduction of the characters to the narrative, the reintroduction of the characters and their maintenance in the narrative, as well as the number of ambiguous character references. The data also indicate that there is a development path in mastering how to refer to characters in an unambiguous way. A statistically significant difference between the three groups of subjects was found in the number of ambiguous character references. This is a clear indication that the children have not reached the same level of pragmatic abilities and sensitivity to the interlocutor’s perspective as adult speakers. Younger children demonstrated a lower success rate in the unambiguous character reference than older children. It is interesting to note that in a small number of cases adult subjects also made mistakes. This can be explained by the complexity of the story chosen for narrative elicitation.